Pipeline

Note: Read recruit-config/user.md for the hiring manager's name, email, and preferences. Do not hardcode any user-specific values.

Config: Read these files before executing — they contain all IDs, tokens, templates, and rules:

• recruit-config/user.md — hiring manager email, Zoom link, office, Slack IDs, noise senders
• recruit-config/ashby.md — source IDs, archive reasons, email style, draft tool rules
• recruit-config/waas.md — WAAS pipeline stages, job IDs, MCP tools` — Slack/Gmail config, source IDs, archive reasons
• recruit-config/CLAUDE.local.md — Slack bot token (gitignored)
• recruit-config/jobs/product-engineer.md — job ID, stage IDs, candidate bar, outreach template, tone rules

Pipeline Status

Show the complete recruiting pipeline — every candidate with activity in the last 21 days.

Procedure

Step 0: WAAS health check

Run health_check() first. If the response is WAAS_API: expired or WAAS_API: error, note "WAAS API unavailable — showing Ashby + Gmail data only" in the summary header and skip WAAS-specific steps.

Step 1: Gather all active candidates

Pull candidates from multiple sources to build the full list:

1. Ashby active applications: First call job_list() to get all open jobs, then application_list(status: "Active") for each job — this is the primary source of truth for who's in the pipeline. Do NOT hardcode a single jobId; The hiring manager hires for multiple roles (Product Engineer, Investment Associate, Design Engineer, Recruiter, etc.).

2. WAAS pipeline: Use the WAAS MCP to pull the full pipeline board per job:

mcp__waas__job_list()                          # discover jobs + stage names
mcp__waas__pipeline_show(job_id: <id>)         # full kanban for each job

pipeline_show returns every stage with its candidates (short_id, name, entered_at, state, needs_response). Call it for each job from job_list.

For applicant-level detail (positions, educations, etc.), use applicant_list(compact: true) or candidate_batch(short_ids) (max 25 per call).

To move candidates between stages: mcp__waas__pipeline_move(job_id: <id>, short_ids: ["abc"], stage_name: "Screen")

3. Gmail inbox: gmail_query_emails(query: "newer_than:21d (from:workatastartup OR from:jackandjill OR subject:'Reaching out from' OR subject:'Introduction:' OR subject:'Intro:' OR subject:'Follow up from YC')") — catch candidates who are in email but may not be in Ashby yet.

4. Deduplicate by candidate name/email across all sources.

Step 2: Run candidate-status for each

For each candidate found, run the full pre-flight checklist from the /candidate-status skill:

• Ashby search + application info
• Gmail inbox + sent (14d)
• Gmail drafts
• Calendar (4 weeks) with responseStatus

GATE — Do not present results until this is complete

For EVERY candidate you will show, verify you have queried ALL of:

1. Ashby stage + application info
2. Gmail inbox (14d) — latest email from candidate
3. Gmail sent (14d) — the hiring manager's last message to candidate
4. Gmail drafts — any pending draft for this candidate
5. Calendar (4 weeks) — upcoming events + responseStatus
6. WAAS status (if candidate has short_id)
7. Ashby feedback — for ALL Tech Screen+ candidates (count them, query all N)

If you have N candidates to present, you must have made N × 7 source checks (minus any that don't apply). If you checked fewer, go back and query the missing ones.

Your output MUST start with a "Sources checked" line listing what you queried with counts. If any source was skipped, state which and why.

Step 3: Present the full table

Group by Ashby stage (highest stage first):

## Pipeline (as of {date})
Checked: Ashby ({N} active across {N} jobs), Gmail inbox (21d), Gmail sent (14d), Drafts, Calendar (4 weeks), WAAS ({N} applicants), Feedback ({N}/{N} Tech Screen+ candidates)
Skipped: [none / list with reason]

### 🔴 URGENT — Action needed now
Candidates where:
- Candidate replied 3+ days ago and the hiring manager hasn't responded
- Invite sent but NOT accepted and event is within 48 hours
- Candidate asked a question that's blocking scheduling
- Post-interview decision overdue

| # | Candidate | Background | Stage | Waiting since | Action Needed |

### 🟡 NEEDS ACTION — Reply or decide
Candidates where:
- Candidate proposed times or confirmed — needs calendar invite
- Post-screen follow-up needed (advancement or rejection)
- Candidate emailed with a question
- Stale 7+ days — re-engage or close out

| # | Candidate | Background | Stage | Waiting since | Action Needed |

---

### Full Pipeline by Stage

### Offer
| # | Candidate | Background | Source | Status |

### Onsite
| # | Candidate | Background | Source | Status |

### Partner Interview
| # | Candidate | Background | Source | Status |

### Technical Screen
| # | Candidate | Background | Source | Status |

### Phone Screen
| # | Candidate | Background | Source | Status |

### Reached Out
| # | Candidate | Background | Source | Status |

### In Consideration
| # | Candidate | Background | Source | Status |

### WAAS Pipeline (by stage)

Use `pipeline_show(job_id)` for each job to get the full kanban view. Present each WAAS stage with its candidates:

### WAAS — Offer
| # | Candidate | Job | Since | Needs Response | Status |

### WAAS — Interview
| # | Candidate | Job | Since | Needs Response | Status |

### WAAS — Screen
| # | Candidate | Job | Since | Needs Response | Status |

### WAAS — Reached Out
| # | Candidate | Job | Since | Needs Response | Status |

### WAAS — In Review
| # | Candidate | Job | Since | Needs Response | Status |

### WAAS — Applied (not yet staged)
| # | Candidate | Job | Since | Needs Response | Status |

Priority section goes first — the 🔴/🟡 tiers are the quick "what do I need to do" view. The full pipeline below gives the complete roster.

Show ALL candidates in the full pipeline — scheduled, waiting, on hold, everything. Candidates that appear in 🔴/🟡 also appear in the stage tables (with a note like "⚠️ needs action").

Filter rules for priority tiers:

• Exclude from 🔴/🟡: Scheduled and accepted (they're coming), ball in candidate's court (the hiring manager's last email was more recent), archived/rejected
• Include in 🔴: 3+ days waiting on the hiring manager, event within 48 hours not accepted, blocking questions
• Include in 🟡: Needs scheduling, needs decision, needs follow-up, stale 7+ days

Within each tier, order by: days waiting (longest first) → pipeline stage (later stages first) → candidate quality signal.

Use all rules from the playbook at .coding-agent-plans/recruiting-assistant.md — especially the pre-flight checklist, responseStatus checking, and "what to say when you don't know" guidelines.